| Name of the Consulting Firm | Mathematica |
|---|---|
| HQ Location | Princeton, New Jersey, United States |
| Size | Mid (1.5k+) |
| Website | https://mathematica.org |
Overview
Mathematica is an employee-owned, evidence-first research firm that applies ‘human-in-the-lead’ AI to public policy. Rather than selling AI for its own sake, it evaluates whether AI tools are trustworthy enough for real policy decisions — and builds the guardrails that let agencies use them responsibly.
AI-Focused Regulatory Work
Mathematica’s pitch is quietly radical in a hype-saturated market: trust in AI isn’t built on promises, it’s built on evidence. Guided by its published AI Principles & Position Statement, the firm structures rigorous evaluations of analytic capability so agencies and foundations know exactly when — and when not — to let a model touch a funding allocation or a program-effectiveness estimate. That evaluation-first posture makes it a natural fit for the kind of careful, defensible AI adoption policymakers increasingly demand.
Notable AI Policy Work
- Built a custom AI tool with a partner that cut the Centers for Medicare & Medicaid Services’ response times to certain hospital questions by 35% — a widely cited example of ‘governed AI’ in practice.
- Developed a prototype cloud-based framework for evaluating large language models on complex, survey-based policy data — testing whether models can produce defensible, policy-relevant estimates before they’re deployed.
- Filed ‘Solidifying American Leadership in AI,’ its formal response to the White House AI Action Plan RFI, drawing on ML work spanning wastewater public-health surveillance, at-risk-student identification, and hospital-admission prediction.
Thought Leadership
- Artificial Intelligence in Action
Mathematica’s ‘AI in Action’ — human-in-the-lead, governed AI for public-sector impact, including a CMS tool that cut response times 35%. - Trusting Your LLM: AI’s Handling of Complex Statistical Data
Mathematica’s framework for evaluating whether LLMs can produce defensible, policy-relevant estimates on complex survey data. - Solidifying American Leadership in AI (AI Action Plan RFI response)
Mathematica’s response to the White House AI Action Plan RFI, drawing on its cross-domain public-sector ML work.
Clients
Strong quant/research culture; good fit